Elevated plasma CXCL12 leads to pain chronicity via positive feedback upregulation of CXCL12/CXCR4 axis in pain synapses

Abstract

Background

Chronic pain poses a clinical challenge due to its associated costly disability and treatment needs. Determining how pain transitions from acute to chronic is crucial for effective management. Upregulation of the chemokine C-X-C motif ligand 12 (CXCL12) in nociceptive pathway is associated with chronic pain. Our previous study has reported that elevated plasma CXCL12 mediates intracerebral neuroinflammation and the comorbidity of cognitive impairment in neuropathic pain, but whether it is also involved in the pathogenesis of pathologic pain has not been investigated.

Methods

Intravenous or intrathecal injection (i.v. or i.t.) of recombinant mouse CXCL12, neutralizing antibody (anti-CXCL12) or AMD3100 [an antagonist of its receptor C-X-C chemokine receptor type 4 (CXCR4)] was used to investigate the role of CXCL12 signaling pathway in pain chronicity. Two behavioral tests were used to examine pain changes. ELISA, immunofluorescence staining, Western blot, quantitative Real Time-PCR and Cytokine array were applied to detect the expressions of different molecules.

Results

We found that increased plasma CXCL12 was positively correlated with pain severity in both chronic pain patients and neuropathic pain model in mice with spared nerve injury (SNI). Neutralizing plasma CXCL12 mitigated SNI-induced hyperalgesia. A single i.v. injection of CXCL12 induced prolonged mechanical hyperalgesia and activation of the nociceptive pathway. Multiple intravenous CXCL12 caused persistent hypersensitivity, enhanced structural plasticity of nociceptors and up-regulation of the CXCL12/CXCR4 axis in the dorsal root ganglion (DRG) and spinal dorsal horn (SDH). However, intrathecal blocking of CXCL12/CXCR4 pathway by CXCL12 antibody or CXCR4 antagonist AMD3100 significantly alleviated CXCL12-induced pain hypersensitivity and pathological changes.

Conclusions

Our study provides strong evidence that a sustained increase in plasma CXCL12 contributes to neuropathic pain through a positive feedback loop that enhances nociceptor plasticity, and suggests that targeting CXCL12/CXCR4 axis in plasma or nociceptive pathways has potential value in regulating pain chronicity.

Highlights

1) Plasma CXCL12 in neuropathic pain states were positively correlated with pain severity.

2) A sustained increase in plasma CXCL12 contributes to pain chronicity.

3) Elevated plasma CXCL12 enhances spinal synaptic structural plasticity of pain through positive feedback.

Introduction

Chronic pain, defined as a debilitating pain lasting at least 3 months or more, impacts about 25% of the global population in all aspects of their lives, including a number of comorbidities such as mood, sleep and/or memory disorders [14, 17]. Neuropathic pain is the most serious category [68], which is caused by various nerve injuries (e.g., thoracic surgery, spinal cord injury), endocrine dysfunctions (e.g., painful peripheral diabetic neuropathy), viral infections (e.g., postherpetic neuralgia), and cancer or neurologic disorders (e.g., multiple sclerosis, stroke, centralized pain syndromes) [56]. It not only causes intense, persistent and devastating pain, but also currently lacks of effective treatment [12]. Although the concept of “transition from acute to chronic pain” is questionable [19], identifying the underlying mechanism remains key to effective management of pain chronicity.

It is generally believed that the pathogenesis of chronic pain mainly related to peripheral and central sensitization [18, 42]. Persistent structural and functional changes (such as increased excitability) in pain pathway, i.e. neuroplasticity (including nociceptor plasticity and synaptic plasticity), are the basis of pain chronicity [24, 60, 62]. Neuronal plasticity and signal transduction in dorsal root ganglion (DRG, the location of primary somatosensory neuron cell bodies) and spinal dorsal horn (SDH, the intermediate processing center of sensory transmission, integration and regulation) are known to be critical for the development and maintenance of neuropathic pain [33]. However, the mechanistic underpinnings of persistent neuroplasticity at the first pain synapse remains remain largely undiscovered.

CXCL12, or stromal cell-derived factor 1 (SDF1), is initially identified in the immune system and is later found in the nervous system [3]. CXCR4, a G-protein-coupled receptor specific to CXCL12, is widely expressed by most cells, including hematopoietic cells, endothelial cells, tumor cells, neurons and stem cells, and is involved in the upstream of the inflammatory pathway due to its chemotaxis to inflammatory cells [45] and the regulation of nervous system function [55]. Growing evidence has demonstrated that both CXCL12 and CXCR4 are upregulated in neuronal and glia cells in peripheral and central nervous systems and play critical roles in the development and maintenance of pathological pain [13, 25, 44, 48, 70, 71, 75, 76, 79]. Therefore, the CXCL12/CXCR4 signaling pathway in the nervous system may be a potential target for the treatment of neuropathic pain [49]. Interestingly, our recent study found elevated plasma CXCL12 in both chronic pain patients and neuropathic pain model of mice with spared nerve injury (SNI), and validated that it is the key to hippocampal neuroinflammation and cognitive impairment [50]. In addition, a large number of clinical studies have shown that plasma CXCL12 is elevated in a variety of acute or chronic diseases, accompanied by varying degrees of secondary pain, such as rheumatoid arthritis [22], knee osteoarthritis (OA) [23], tuberculosis infection and disease [36], postmenopausal osteoporosis [78], IgG4-related disease [8], pre-eclampsia [67], squamous cell carcinoma [37], chronic kidney disease [53], myocardial infarction [35, 81]. Yang et al. [78] also demonstrated that increased plasma CXCL12 levels in postmenopausal osteoporosis is significantly correlated to pain severity established by visual analogue scale, as well as plasma TNF-α levels. However, whether elevated plasma CXCL12 is associated with the onset and maintenance of pain chronicity and the cause of the elevation is still unknown.

In this study, we used clinical chronic pain patients and SNI neuropathic pain mice to further analyze whether plasma CXCL12 was correlated with pain severity, and focused on whether and how plasma CXCL12 elevation is involved in the progression of SNI neuropathic pain. In addition, the reasons of its increase were also discussed. Our findings demonstrated that elevated plasma CXCL12 was positively associated with chronic pain severity. Mechanically, we disclosed that a sustained increase in plasma CXCL12 enhanced the plasticity of nociceptive receptors in spinal dorsal horn through the CXCL12/CXCR4 positive feedback loop, thereby contributing to the chronicity of pain.

Materials and methods

Animals

A total of 197 C57BL/6 mice (25 females and 172 males) aged 7 ~ 10 weeks (Laboratory Animal Center, Sun Yat-sen University, China) were used in 4 cohort experiments (Table 1). After we found that the gender difference was no significant in Figs. 1B, C and 2 with a 50/50 split between males and females, C57BL/6 male mice were used in follow-up experiments. Animals were housed in animal facilities with a 12/12 light dark cycle, 23 ± 1℃, and ad libitum access to food and water. Mice were pre-habituated for one week prior to experimental and behavior testing. All the behavioral tests were performed by two experimenters, one blinded to treatments. All animal protocols and experimental procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of Sun Yat-sen University (No. SYXK (yue) 2022–0081).

Table 1 Details of C57BL/6 mice used in 4 cohort experiments

Fig. 1

Elevated plasma CXCL12 is positively associated with chronic pain.A Scatterplots indicates a positive correlation between plasma CXCL12 concentration and Numerical Rating Scale (NRS) scores in healthy controls (n = 40) and patients with chronic pain or subgroups (n = 30) and subgroups. Note: female (F), male (M), chronic pain patients (P), healthy controls (C), NP = neuropathic pain + postherpetic neuralgia (PHN) + trigeminal neuralgia (TN), osteoarthritis (OA), chronic back pain (CBP), complex regional pain syndrome (CRPS). The data of plasma CXCL12 concentration comes from our previous study [50]. B ELISA result shows plasma continued to increase after spared nerve injury (SNI). Sham (Sh) group includes 3, 7, 21 days after sham surgery. S3d, S7d, S21d: 3, 7, 21 days after SNI. **P < 0.01, ***P < 0.001 vs. Sh group; # P< 0.05, ## P < 0.01 vs. S1d group. C Scatterplot analysis indicates that plasma CXCL12 concentration in (B) was negatively correlated with paw withdrawal threshold (PWT) in mice

Fig. 2

CXCL12 neutralizing antibody relieves SNI-induced hyperalgesia. A Experimental diagram of multiple intravenous injection (i.v.) of CXCL12 neutralizing antibody in SNI model mice. The black lines represent mechanical pain behavioral tests. The blue lines show injection of anti-CXCL12 antibody (ab9797, Abcam, 20 ng/200 μl, once a day for 8 days) and the red line indicates sham surgery or SNI time point. B Continuous i.v. injection of anti-CXCL12 antibody partially alleviated ipsilateral mechanical allodynia caused by SNI (half male in each group). *P < 0.05, **P < 0.01, ***P < 0.001 vs. intravenous vehicle and sham control group (Vehi + anti-CXCL12) # P< 0.05, ## P < 0.01, ### P < 0.001, #### P < 0.0001 vs. intravenous vehicle and SNI group (Vehi + SNI)

Surgery induction (SNI induced pain model)

Mice were anaesthetized with 2% isoflurane during surgery. The SNI model has been described in the previous protocol [16, 43]. Briefly, the left sciatic nerve and its three branches (peroneal, tibial and sural nerves) were exposed and identified. Peroneal and tibial nerves were isolated and ligated with 7–0 sutures. The stump about 0.2 mm outside the ligation site was cut off, and then the muscle and skin were closed layer by layer. In the sham operation group, only 3 branches of the sciatic nerve in the popliteal fossa were exposed without nerve damage. After recovery from anesthesia, the animals were housed under normal conditions.

Drug administration

According to our previous study [50], plasma was neutralized using 20 ng of anti-CXCL12 antibody (Cat# ab9797, Abcam, 200 μl) via a caudal vein injection once every other day from 30 min before and after sham or SNI surgery to 9 d after surgery. Control mice were injected with an equal amount of 0.1% bovine serum albumin (BSA) in the same way. In addition, anti-CXCL12 antibodies (200 ng, 10 μl) or the antagonist of its receptor CXCR4 (AMD3100, Cat# A5602, Sigma-Aldrich, 10 μg, 10 μl) was i.t. injected to neutralize CXCL12 or inhibit CXCR4 in pain pathways such as DRGs and SDHs.

The recombinant mouse CXCL12 protein (Cat#P4371, Abnova) was prepared into a 100 μg/ml storage solution and stored in a -80 °C refrigerator for use, and then diluted to the working concentration in 0.1% BSA) normal saline before administration. According to our previous study [50], i.v. injection of CXCL12 at 1 ng/ml or 2.5 ng/ml (200 μl) could mimic plasma pathological concentration in SNI model. We also selected a lower concentration of 0.1 ng/ml CXCL12, and a sterile saline volume containing 0.1% BSA as vehicle (Vehi) control. The number of injections n is represented by n × .

Mechanical sensitivity (von Frey) test

Accordance to the previous studies [9, 40], paw mechanical threshold (PWT) of mice was determined by the up-down method via von-Frey filaments (0.04–1.4 g, North Coast medical, USA). In short, the mice were placed in separate transparent chambers 0.5 h before test to ensure quiet but not sleepiness. One filament was applied vertically to the center of the plantar surface of the mouse’s hind paw until it formed an S-shaped bend, and stopped after 4–6 s stimulation. A positive response indicated foot withdrawal or licking behavior, while no response indicated a negative reaction. The initial stimulus intensity was 0.04 g and the maximum was 1.4 g. Each stimulus was repeated 5 times with an interval of 30 s. Mice with a basal threshold (BS) of less than 0.6 g before surgery or treatment were excluded.

Acetone test

The mice were placed in separate compartments 0.5 h before the test and kept quiet and awake. Acetone (HB02, Guangzhou Chemical Reagent Factory) was evenly sprayed in the middle plantar of the hind paw with a plastic syringe. The response latency and behavioral score were recorded in 3 repetitions at intervals of no less than 1 min (1 point for foot lift/shaking, 2 points for foot licking, and 1 point for behavioral continuity) referred to previous literature [40].

Plasma sample collection and plasma CXCL12 ELISA assay

To examine the changes of plasma CXCL12 after mouse SNI model, blood samples were collected 1, 3, 7 and 21 days after sham or SNI operation and plasma were extracted for ELISA detection. As follows, the mice were anesthetized with 2% isoflurane and fixed in a supine position. The chest wall was cut from xiphoid process to subclavicle, and the pericardium was torn to expose the heart. Then the right atrium was intubated with a 1 ml syringe, and 0.5–0.6 ml returned blood was drawn and placed into the EDTA anticoagulant tube. The blood samples were centrifuged at 3000 rpm at 4℃ for 15 min, and then the supernatant was subpackaged and stored at -80℃.

CXCL12 levels were measured in mouse plasma samples according to the manufacturer’s instructions of ELISA kit (ab100741, Abcam). In brief, the samples or standard were added to a 96-well plate and incubated for 90 min at room temperature (RT). Finally, the absorbance value was measured by the enzyme labeling method at 450 nm wavelength, and the standard curve was drawn to calculate the sample concentration.

Immunofluorescence staining (IF) and analysis

Mice were subjected to deep anesthesia by 10% Urethane, followed by transcardial perfusion with 0.01 M cold PBS and then 4% paraformaldehyde (PFA). L4 ~ 6 DRGs and SDHs were harvested and post-fixed for 4 h ~ 1d. After dehydrating with 30% sucrose in PBS at 4℃ for 2 days, all tissues were sliced into 16 μm (DRGs) or 18 μm (SDHs) sections by using a cryotome (CM3050S, Lecia, Germany), and transferred on to the Superfrost Plus Microscope slides (FD Neuron Technologies, Inc, USA). After RT blockage in 5% donkey serum in 0.3% Triton X-100 (Sigma) for 1 h, the slices were incubated with primary antibodies overnight at 4℃. The following antibodies were used: rabbit anti-p-ERK (4370, CST, 1:500), goat-anti-CGRP antibody (3600, Abcam, 1:1000), mouse anti-GFAP antibody (3670, CST, 1:500), rabbit anti-p-P38 (4511, CST, 1:400), goat anti-Iba1 (5076, Abcam, 1:1000), mouse anti-VGLUT2 antibody (MA5-27613, Invitrogen, 1:100), mouse anti-NeuN (104224, Abcam, 1:1000), rabbit anti-SDF1/CXCL12 antibody (18919, 1:500, Abcam), rabbit anti-CXCR4 antibody (100–74396, Novus Biologicals, 1:500), rabbit anti-c-FOS antibody (2250S, CST, 1:500). Following 3 washes with PBS, the slices were then incubated with secondary antibodies (Alex Flour 488, 555, 647; Life technologies, USA) at RT and washed for 3 times with PBS. The slices were protected by Antifade Mounting Medium with DAPI (P0131, Beyotime, China) under coverslips and fluorescent images were obtained with a fluorescence microscope (EVOS FL, Thermo Fisher Scientific, USA).

There were 3 to 4 mouse samples in each group, 4 slices with strong positive signal in the limited nuclear range were blindly selected for statistics. The total number of positive immunoreactive cells (p-ERK, p-P38, Iba1, c-FOS, CXCL12, CXCR4) and the relative fluorescence density of other signaling in each section were calculated by ImageJ software (National Institutes of Health, Bethesda, MD) to show the protein expression. The intensity in sample from sham of Vehi control was set as baseline 1.

Western Blot (WB)

After deep anesthesia with 10% Urethane, mice were perfused intracardially with 0.01 M PBS (4℃, pH = 7.4). L4 ~ 6 DRGs and SDHs were carefully dissected, homogenized and sonicated in a lysis buffer containing a protease inhibitor cocktail (P1045, Beyotime Bio-technology). The lysates were centrifuged at 12 000 rpm for 20 min and the concentration of the supernatants were determined by BCA Protein Assay Kit (23327, Thermo Fisher Technology, USA). 15 μg/20 μl sample was placed on the bottom of 15% SDS-PAGE sample well for electrophoretic separation, and then wet-transferred onto 0.22 μm PVDF membrane. The membrane was blocked with 2% BSA in TBST (Tris-buffered saline, 0.1% Tween 20) at RT for 1 h, and then incubated at 4℃ overnight with primary antibody: rabbit anti-p-ERK (4370, CST, 1:1000); rabbit polyclonal to SDF1 (18919, Abcam, 1:1000,), rabbit polyclonal to CXCR4 (100–74396, Novus Biologicals, 1:1000) and mouse monoclonal antibody to GAPDH (60004–1-1 g, Proteintech, 1:50000). After incubation with horseradish peroxidase labeled anti-rabbit secondary antibody (1:10000, Abcam) or anti-mouse secondary antibody (1:10000, Abcam) at RT for 1 h, protein band was detected by ECL (BL523B, Biosharp) chemiluminescence solution and captured with TANON-5200 chemiluminescence imaging system (Tanon Technology). Integrated gray value was analyzed by ImageJ. The ration of average optical density of target protein/GAPDH in Vehi/sham group was standardized as 1.

RNA isolation and quantitative Real Time-PCR (qRT-PCR)

According to our previous study, L4 ~ 6 DRGs and SDHs were excised and homogenized by RNAzol® RT (RN 190–200, MRC, China). RNA was isolated by TriZol/chloroform extraction, and RNA concentration and purity were detected using Nano drop 2000 (Thermo Fisher Scientific, USA). qRT-PCR was performed using CFX 96 touch (C1000™, Bio-rad, USA) and 2 × Master qPCR Mix SYBR green I (TES201, TSINGKE, China) in following conditions: 95 °C for 30 s; 40 cycles of 95 °C for 5 s, 60 °C for 30 s, and finally melting analysis. The data analysis of mRNA expression was normalized to the internal control Gapdh and quantified by the 2^–ΔΔCt method. The ratios of sham/con/Vehi group were set as baseline 1. All primers used for qRT-PCR are listed in Table 2.

Table 2 Primer for RT-qPCR in tissues

Cytokine array for sciatic nerve and DRG samples

Mice were transcardially perfused with cold PBS at 7 d after sham SNI surgery. The proximal stump of the ipsilateral sciatic nerve and L4 ~ 6 DRGs were dissociated and homogenized in a lysate buffer containing protease and phosphatase inhibitors. After the protein concentration was determined by BCA protein assay (23327, Thermo Fisher Technology, USA), the cytokine expression in the samples was analyzed by Proteome Profiler Mouse Cytokine Array Kit, Panel A (#ARY006, R&D). Each reaction was performed according to the manufacturer’s protocol of using 100 μg proteins collected from 1 sample. Protein samples were incubated at 4 °C for 24 h with mouse cytokine arrays pre-coated with 40 cytokines/chemokines. The rest of the procedure was the same as WB.

Statistical analysis

Data were presented as mean ± SEM. The Pearson correlation coefficients were calculated to determine the relationship between human plasma CXCL12 (pg/ml) and NRS Score or mice plasma CXCL12 (pg/ml) and PWT (g). When data in groups were normally distributed with equals variance, unpaired T-test was performed between 2 groups. Behavior results containing time and groups two factors were tested by two-way repeated-measure ANOVA (followed by Turkey’s multiple comparison test). Other values of each group were tested using one-way ANOVA, followed by Turkey’s multiple comparison test. Statistical analysis was calculated using Graphpad Prism 9.0.0 (GraphPad Software, LLC, CA, USA). The significance threshold was set at P < 0.05.

Results

Elevated plasma CXCL12 is positively correlated with chronic pain severity

To understand the relationship between plasma CXCL12 and pain severity in some common chronic pain patients, we first reanalyzed the clinical data on plasma CXCL12 and cognitive impairment in healthy controls (n = 40) and chronic neuropathic pain patients (n = 30) from our recent study [50]. The 30 patients with chronic pain were mainly divided into the following subgroups: neuropathic pain (NP, n = 16), osteoarthritis (OA, n = 9), complex regional pain syndrome (CRPS, n = 9), chronic back pain (CBP, n = 11). The pain severity was assessed by Numerical Rating Scale (NRS). After Spearman rank correlation analysis, we found that plasma CXCL12 concentration was positively correlated with NRS score in healthy controls and in all chronic pain patients (r = -0.6100, R² = 3.721, P < 0.0001, Fig. 1A). This is consistent with a previous report on chronic pain patients with postmenopausal osteoporosis [78]. Among these subgroups, the association was moderate in patients with neuropathic pain (NP). This means that it is feasible to explore the role and mechanism of plasma CXCL12 elevation in chronic pain represented by neuropathic pain. So next we also re-examined the changes of plasma CXCL12 in mouse SNI models at different times by ELISA assay. In agreement with our previous study [50], plasma CXCL12 was elevated from SNI 3d and persisted for at least 21 d (Fig. 1B, from Sham group 23.42 ± 4.12 pg/ml to SNI 3d 45.53 ± 2.65 pg/ml and S21d 69.49 ± 8.00 pg/ml), and no significant gender difference was observed. Correlation analysis showed that plasma CXCL12 concentration in SNI mice was negatively correlated with PWTs (Fig. 1C), suggesting that plasma CXCL12 may play an important role in the chronicity of pain.

CXCL12 neutralization mitigates SNI-induced pain hypersensitivity

To assess the impact of elevated plasma CXCL12 on chronic pain, multiple intravenous injections of anti-CXCL12 antibody (20 ng/200 μl, 1 d before to 8 d after surgery, once a day) were administered into the tail veins of sham or SNI model groups (Fig. 2A). Anti-CXCL12 antibody had no significant effect on PWT in the sham operation group. However, sustained mechanical hypersensitivity induced by SNI was partially alleviated by multiple administration of CXCL12 neutralizing antibodies compared with solvent control (Fig. 2B). Taken together, these results demonstrated that the increase of plasma CXCL12 is involved in the progression of SNI-induced neuropathic pain.

Single intravenous CXCL12 induces mechanical hyperalgesia and enhances neuronal excitability of primary pain pathways

To further clarify the relationship between sustained increase of plasma CXCL12 and the pathologic pain, we then gave naive mice a single intravenous infusion of exogenous CXCL12 through the tail vein to simulate the elevated plasma CXCL12 and observed its effect on mechanical pain threshold. According to our previous study on mouse SNI model [50] and the above Elisa results (Fig. 1C, plasma CXCL12 < 100 pg/ml), we selected different pathologic concentrations (0.3, 1.0, 2.5 ng/ml, 200 μl) of CXCL12 for i.v. administration in naïve mice. As expected, 0.3 ng/ml CXCL12 had no effect on PWTs compared with the 0.1% BSA Vehi group (Fig. 3A), but two high concentrations of CXCL12 significantly reduced bilateral PWTs at 2 h after injection, progressed to day 3, and then gradually returned to normal. There was no significant difference between these two concentrations, suggesting that 1.0 ng/ml might be the optimal concentration. Subsequently, to confirm the results of pain behavioral responses, we also observed the expressions of several pain signaling molecules, such as phospho-extracellular signal-regulated kinase (p-ERK), phospho-p38 mitogen-activated protein kinase (p-P38), Calcitonin Gene-Related Peptide (CGRP) and astrocyte marker GFAP [26, 30, 32, 34, 51, 69] in the pain pathway by immunofluorescence staining at 1 d after injection. Compared with the Vehi group, 1 ng/ml CXCL12 not only caused dramatic increases in the expressions of p-ERK, p-P38 and CGRP in L4 ~ 6 DRGs (Fig. 3B-E), but also upregulated the expressions of CGRP and GFAP in L4 ~ 6 DRGs and the superficial layer of SDH. In addition, the upregulated p-P38⁺ cells were mainly surrounded by a large number of non-neuronal small cells deeply staining DAPI, and a small part is co-stained with macrophage marker Iba1(Fig. 3B), suggesting that upregulated p-P38 in DRGs is mainly expressed in neurons. In contrast, p-P38 in SDH of the Vehi group was expressed in Iba1-labeled microglia (Fig. 3F, G). Interestingly, a single injection of CXCL12 induced upregulation of p-P38 in SDH without a significant increase in the number of microglia. The upregulated p-P38 was expressed in Iba1⁻ cells, which may be superficial projection neurons. Taken together, these results demonstrate th6at transient pathologic elevation of plasma CXCL12 leads to mechanical hypersensitivity and activation of pain pathways in DRG and SDH.

Fig. 3

Intravenous injection of CXCL12 induces prolonged mechanical hypersensitivity and activation of nociceptive pain pathway. A The effects of a single intravenous injection (i.v., 200 μl) of vehicle (Vehi) or CXCL12 (0.3, 1.0, 2.5 ng/ml) on PWT in male naïve mice. h: hour(s), d: day(s). *P < 0.05, **P < 0.01, ***P < 0.001 vs. Vehi group; #P < 0.05 vs. 0.3 ng/ml CXCL12 group. B-G Representative images and quantification showing the changes of p-ERK, p-P38, CGRP and GFAP signals in DRGs and/or spinal dorsal horn (SDH at 1 d after i.v. injection of 1 ng/ml CXCL12 (n = 4 mice/group, 3 ~ 4 sections/mouse). The expression of p-P38 in the dashed line frame is shown with nuclear (DAPI labeled) /Iba1 co-staining. The yellow arrows in B and F indicate co-immunostaining, while the red arrows show only p-p38 signals are expressed. *P < 0.05, **P < 0.01, ***P < 0.001 compared to Vehi group

Multiple intravenous CXCL12 results in pain chronicity and structural enhancement of nociceptor plasticity

Given the above results showing that 1.0 ng/ml CXCL12 (200 μl) induced hypersensitivity for 3 days (Fig. 3A), we then employed multiple injections of CXCL12 (4 injections every other day) to simulate the long-term effect of sustained increases in plasma CXCL12 on pain behavioral response and the structure of pain synapses (Fig. 4A). 4 injections of 0.1% BSA in the saline vehicle control group (Vehi 4 ×) had little effect on the pain threshold, indicating that multiple injections did not affect pain response behavior (Fig. 4B). PWTs decrease after single injection of CXCL12 (1 ×) and recovered gradually on day 4 without no statistical difference compared with Vehi group. Surprisingly, 4 intravenous injections of CXCL12 (4 ×) caused pain hypersensitivity over 2 weeks. In addition, acetone test results showed that single or multiple CXCL12 reduced the latency of bilateral cold withdrawal and [40] increased the cold pain score in varying degrees and time-dependent manner (Fig. 4C). However, in the Hargreaves test, intravenous CXCL12 had no significant effect on the thermal withdrawal threshold compared to the control group (data not shown).

Fig. 4

Multiple intravenous injections of CXCL12 cause pain hypersensitivity. A Experimental flow chart. A single (1 × , 200 μl) or 4 consecutive (4 × , on every other day) CXCL12 (1.0 ng/ml) or Vehi was intravenously injected in naïve male mice. B Changes of PWT showed that intravenous CXCL12 induced mechanical allodynia in different degrees. C Acetone test indicated that intravenous CXCL12 caused cold hyperalgesia. *P < 0.05, **P < 0.01, ***P < 0.001 vs. Vehi group; # P < 0.05, ## P < 0.01, ### P < 0.001 vs. CXCL12 1 × group

Since pathologic pain can cause continuous changes in the structure and function of pain transduction pathways, we also collected samples at different time points and observed the changes of molecules CGRP, which are related to pain pathways through western blot and immunostaining. Compared with the Vehi group, multiple intravenous administration of CXCL12 resulted in upregulation of p-ERK in DRG at 3d and SDH at 7d (Fig. 5A). The results of p-ERK in SDH were confirmed by immunofluorescence staining (Fig. 5B, D). As expected, multiple intravenous CXCL12 also resulted in significant overexpression of CGRP⁺ terminal and VGLUT2 (Fig. 5C, D), which is crucial in mediating nociceptive transmission in the superficial layer of SDH [73]. Although these signals were more or less upregulated in the 1.0 ng/ml CXCL12 1 × 7d or 14d and 4 × 7d groups compared to the Vehi group, only the CXCL12 4 × 14d group had an overwhelming increase (Fig. 4F, G). Combining behavioral and molecular biological results, we conclude that a sustained pathological increase in plasma CXCL12 does lead to the chronicity of pain by causing a long-term enhancement of molecular plasticity in nociceptive synapses.

Fig. 5

Multiple intravenous CXCL12 induced sustained enhancement of spinal synaptic plasticity. A Western blot results showing the changes of p-ERK expression in the ipsilateral L4 ~ 6 DRGs and SDH at 3 or 7 d after Vehi or CXCL12. *P < 0.05, **P < 0.01, ***P < 0.001 between groups. B-D Immunostaining results showing the expressions of p-ERK, CGRP and VGLUT2 in SDH at 7 or 14 days after first (1st) treatment (n = 3 mice/group, 3 ~ 4 sections/mouse). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 vs. Vehi group; # P < 0.05 vs. CXCL12 1 × 7d group; % P< 0.05, %%% P < 0.001 vs. CXCL12 4 × 7d group; & P < 0.05, &&& P < 0.001 vs. CXCL12 1 × 14d group

The increase of plasma CXCL12 induced the upregulation of CXCL12/CXCR4 signaling in DRG and SDH

Under certain pathologic pain conditions, CXCL12 and its receptor CXCR4 are upregulated in DRG and SDH, primarily in neurons, but also in glial cells [5, 44, 54, 61], and play a key role in the development and maintenance of pathologic pain [49]. Therefore, we used qRT-PCR and IF to observe whether plasma CXCL12 affected the expression of this signaling axis in the pain transmission pathway. Compared with Vehi and CXCL12 1 × groups, the expression of CXCL12 and CXCR4 mRNA in DRGs was significantly increased at day 3 after CXCL12 2 × administration (Fig. 6A). But the changes in the SDH are not obvious. Western blot data further demonstrated that CXCL12 and CXCR4 proteins were upregulated not only in DRG but also in SDH at 3d or 7d after multiple CXCL12 treatment compared to Vehi and single CXCL12 groups (Fig. 6B, C). The IF results clearly showed that CXCR4 protein expression was markedly upregulated in L5 DRGs 1 or 7 days after 1 × or 4 × intravenous administration compared to Vehi group, but not 7 days after a single dose (Fig. 6D, E). Double staining showed that CXCL12 was upregulated in NeuN-labeled neurons but not in GFAP-labeled satellite glial cells. Additionally, CXCL12 was increased in SDH at day 14 after CXCL12 1 × treatment, but overwhelmingly upregulated in SDH at 14 days after CXCL12 4 × treatment (Fig. 6F, G). Immunofluorescence double staining demonstrated that CXCL12 was mainly expressed in neurons (Fig. 6H marked by NeuN) and microglia (Iba1), but not in astrocytes (GFAP). CXCR4 increased significantly on day 7 of CXCL12 1 × or 4 × processing (Fig. 6I, J). However, like CXCL12, CXCR4 expression was most obvious at day 14 after CXCL12 4 × treatment. Additionally, CXCR4 was expressed in neurons but not in microglia and astrocytes in SDH (Fig. 6K). These findings indicate that a sustained increase in plasma CXCL12 can upregulate the expression of CXCL12/CXCR4 signaling axis in DRG and SDH.

Fig. 6

Elevated plasma CXCL12 leads to the upregulations of CXCL12/CXCR4 axis in DRG and SDH. A RT-qPCR analysis showing the mRNA expressions of Cxcl12 and its receptor Cxcr4 in DRG and SDH. *P < 0.05, **P < 0.001 vs. Vehi group. B, C Western blot results showing the upregulation of CXCL12 and CXCR4 in the ipsilateral L4 ~ 6 DRGs and SDH at 3 or 7 d after Vehi or CXCL12. *P < 0.05, **P < 0.01, ***P < 0.001 between groups. D-K Immunofluorescence results demonstrated that the expression changes of CXCL12 (D-F) and CXCR4 (G-I) in L5 DRG or SDH and their co-staining with neurons (marked by NeuN), microglia (Iba1) and astrocytes (GFAP) (n = 3 mice/group, 3 ~ 4 sections/mouse). The yellow arrows in H and K indicate co-immunostaining. *P < 0.05, **P < 0.01, ****P < 0.0001 compared to Vehi group; # P < 0.05, #### P < 0.0001 vs. CXCL12 1 × 7d group; $$$$ P < 0.0001 vs. CXCL12 1 × 14d group, &&&& P < 0.0001 vs. CXCL12 4 × 7d group

Intrathecal blocking of CXCL12/CXCR4 axis mitigates intravenous CXCL12-induced hypersensitivity by inhibiting the positive feedback loop

To determine whether plasma CXCL12 causes chronic pain through positive feedback acting on the nociceptive pathway, we blocked this axis by intrathecal injection (i.t.) of anti-CXCL12 neutralizing antibody or CXCR4 inhibitor AMD3100 30 min before each i.v. injection of CXCL12. We found that both treatments did provide partial relief from the mechanical allodynia caused by multiple i.v. CXCL12 injections (Fig. 7A). The increase in c-FOS and upregulations of CXCL12, CXCR4, and p-P38 in SDH induced by multiple i.v. CXCL12 were also prevented by pre-intrathecal administration of anti-CXCL12 antibody or AMD3100 (Fig. 7B-E). However, the inhibitory effect of AMD3100 on the up-regulation of CXCL12/CXCR4 axis was stronger than that of anti-CXCL12 antibody. Taken together, these results suggest that positive feedback of the CXCL12/CXCR4 axis is involved in the pathological changes of chronic pain and nociceptive pathways induced by plasma CXCL12.

Fig. 7

Intrathecal blocking of CXCL12/CXCR4 axis alleviates intravenous CXCL12 induced-pain hypersensitivity, activation of pain pathways. A Effects of intrathecal injections (i.t., 10 μl, black arrows) of anti-CXCL12 antibodies (200 ng) or CXCR4 inhibitor AMD3100 (10 μg) 30 min before i.v. CXCL12 4 × (1.0 ng/ml, 200 μl, red arrows) on PWT in male naïve mice. *P < 0.05, **P < 0.01, ***P < 0.001 vs. Vehi group; # P < 0.05, ## P < 0.01, ### P < 0.001 vs. CXCL12 1 × group, two-way RM ANOVA (followed by Tukey’s). B-E Immunostaining data showing the expressions of SDH c-FOS (B, C), CXCL12, CXCR4 and p-P38 (D, E) in different groups at 14d after treatment (n = 4 mice/group, 3 ~ 4 sections/mouse). **P < 0.01, ****P < 0.0001 compared to Vehi group, ## P < 0.01, #### P < 0.0001 vs. i.v. CXCL12 group

Damaged nerve stump and DRGs are possible causes of elevated plasma CXCL12 in SNI model

Finally, we explored the possible causes of elevated blood CXCL12 in neuropathic pain from the SNI model. We first detected the expression changes of CXCL12 and its receptor CXCR4 in the ipsilateral SDH and DRG at 7 days after SNI. PCR and WB results showed that the mRNA and protein of CXCL12 and CXCR4 were significantly increased in these two sites to varying degrees (Fig. 8A-C), which were consistent with the results of a previous study on rat SNI model [2]. In addition, since the injured ipsilateral sciatic nerve stumps and DRGs are the initial sites of SNI surgery, we used R&D Systems Mouse Cytokine Array Panel A (#ARY0006) to observe the effect of SNI on cytokine expression in these two tissue homogenates at 7 d after surgery. Compared with sham group (Fig. 8D), SNI sciatic nerve stump exhibited a large number of significantly upregulated pro-inflammatory cytokines (IL-1β, IL-1α, TNF-α, IFN-γ, IL-17, IL-23), chemokines (JE/CCL2, CCL3, CXCL1, CXCL2, CXCL9), complement fragment (C5/C5a), colony-stimulating factors (M-CSF, GM-CSF, G-CSF), adhesion molecules (sICAM1), protease (TIMP-1) and anti-inflammatory factors (IL-1ra, IL-4, IL-10, IL-13), i.e. inflammatory storm. In contrast, the change in DRGs is not as dramatic. However, some molecules such as JE/CCL2, M-CSF, and TIMP-1 also showed significant increases (Fig. 8E). Surprisingly, no chemokine CXCL12 was detected in either ipsilateral sciatic nerve stump or DRG in either sham or SNI group. That is, the increase in plasma CXCL12 may not be directly synthesized and released into the blood through these two sites that are directly associated with SNI damage.

Fig. 8

Changes of CXCL12 signaling in ipsilateral neural pathways and blood following SNI. A-C Expression changes of CXCL12 and its receptor CXCR4 mRNA and protein in the ipsilateral SDH and DRG of SNI. *P < 0.05, ** P < 0.01, vs. Sh group. D, E Cytokine array results showing changes of cytokines and chemokines in the ipsilateral sciatic nerves and L4 ~ 6 DRGs on day 7 after sham and SNI groups. There was no difference in the positive control protein levels (white boxes) between the two groups. Ipsi: ipsilateral, n.: nerve. F Heat map of RNA-Seq data (GEO: GSE111216) from four sites in mouse models of SNI-neuropathic pain and CFA-inflammatory pain [59]. Nine 18-week-old BALB/c female mice were randomly assigned into 3 groups (n = 3 mice/group): control (CON, naive), SNI for modelling neuropathic pain, and Inflammatory pain induced with CFA. The color bars represent the gene expression values of each group compared to the control group at the same site. *P < 0.05, ** P < 0.01, **** P < 0.0001. SC: spinal cord

To gain a detailed understanding of the origin of plasma CXCL12 in SNI neuropathic pain, we reanalyzed the differential gene expressions (DGEs) of inflammatory factors at four pain-related sites in neuropathic and inflammatory pain based on the previous study of next-generation deep RNA-Seq data [59]. mRNA samples were extracted from 3 groups: control (CON), SNI neuropathic pain model on day 7 (3 mice/group, 18-week-old BALB/c female) and complete Freund’s adjuvant (CFA)-induced inflammatory pain on day 3, and 4 tissues: whole blood, DRG (L3-L5), lumbar spinal cord (SC) and whole brain. We found that Cxcl12 mRNA tended to increase (but not statistically significant) in the SNI blood samples compared to the CON group, but not in the CFA inflammatory pain model (Fig. 8F). Shockingly, there were no significant changes in the ipsilateral DRG, SDH, and Cxcl12 mRNA in the brain in the SNI and CFA models. However, there was a significant increase in Csf1, Ccl2 mRNA in SNI 7d in DRG, which was consistent with a previous study [20] and our SNI 7d cytokine array results (Fig. 8D). Most surprisingly, the levels of Il1b and Csf1 mRNA in blood after SNI were significantly higher than those in CFA and sham groups, and Cxcr4, Tnf and Icam1 also tended to increase. This analysis suggests that SNI-induced neuropathic pain does have a significant circulating immune response compared to the CFA-induced inflammatory pain. These changes in damaged nerve stump and DRGs, as well as circulating immune response, may be the cause of increased plasma CXCL12.

Discussion

Numerous previous studies have focused on the role of CXCL12 in the nervous system in the development of neuropathic pain. In contrast, the present work expands the understanding of the important role of extra-nervous CXCL12 in the pathological process of chronic neuropathic pain. The innovations of this paper are mainly in three aspects: 1) we provide evidence that elevated plasma CXCL12 is positively correlated with chronic pain severity in patients and SNI mice. 2) Mouse studies have revealed that a sustained increase in plasma CXCL12 mediates pain chronicity through the positive feedback loop of CXCL12/CXCR4 axis, thereby enhancing the plasticity of spinal pain synapses. 3) SNI induced a dramatic inflammatory immune response in damaged nerve stump and DRGs, as well as in circulation, which may be responsible for the elevated plasma CXCL12. Therefore, targeting CXCL12/CXCR4 axis in plasma or nociceptive pathways has potential value in the treatment of neuropathic pain.

Pathological roles of plasma CXCL12 in pain chronicity

In this study, we found that elevated plasma CXCL12 in clinical cases and neuropathic pain model mice was positively associated with chronic pain severity (Fig. 1A-C), which is consistent with previous studies of postmenopausal osteoporosis [78]. More interestingly, multiple i.v. administration of CXCL12 antibodies largely relieved allodynia induced by SNI (Fig. 2A, B), which in part suggests that an increase of plasma CXCL12 is necessary to cause neuropathic pain. Additionally, a single i.v. injection of exogenous CXCL12 at pathological concentrations caused mechanical hypersensitivity for up to 3 d (Fig. 3A). Furthermore, 4 × CXCL12 injected every other day resulted in pain hypersensitivity for up to 2 weeks (Fig. 4). To sum up, our findings reveal for the first time that sustained elevation of plasma CXCL12 contributes to the chronicity of pain.

Elevated plasma CXCL12 results in structural and functional potentiation of nociceptor plasticity

Previous studies have shown that the underlying mechanism of neuropathic pain mediated by up-regulated CXCL12 in the nervous system mainly involves two aspects of neurons and glia: directly enhancing the excitability of DRG [5, 28, 54, 77] and SDH [41] neurons, promoting inflammation [13, 38], desensitizing mu- and delta-opioid receptors [72] and inhibiting leukocyte-derived endogenous opioid secretion [74]. In this study, our results showed that i.v. injection of CXCL12 for 1 d resulted in sharp increases in the pain signaling molecules p-ERK, p-P38 and CGRP in DRG (Fig. 3B-E) and CGRP, GFAP and p-P38 in superficial SDH (Fig. 2D-G), while no significant change in microglia (marked by Iba1). In addition, continuous i.v. injection of CXCL12 induced long-term enhancement of neuronal excitability and nociceptor plasticity (Fig. 5). These data not only further confirm that plasma CXCL12 causes pathologic pain, but also suggest that its mechanism is related to structural and functional enhancement of nociceptor plasticity.

The mechanism of plasma CXCL12 in pain chronicity

Studies have shown that a single intraplantar injection of CXCL12 (200 ng/10 μl) was sufficient to induce acute mechanical hyperalgesia in rats from 0.5 h to 1 d [76] and a single intrathecal injection of CXCL12 (250 ng) decreased PWT in rats from 1 h to 3 d [44]. Our study demonstrated that intravenous injection of CXCL12 (0.2 or 0.5 ng, 200 μl) significantly reduced PWT in naive mice from 2 h to 3 d. Interestingly, intravenous administration apparently triggered mechanical hypersensitivity for a slightly longer than intraplantar and intrathecal injection, and the effect duration was comparable to that of intrathecal injection (Fig. 3A). This suggests that it takes some time for plasma CXCL12 to induce a nervous system response.

There is evidence that most PNS and CNS have low permeability between blood and nerve tissue due to the presence of blood–brain barrier (BBB) and blood-nerve barrier (BNB), but intact DRGs show high permeability through their loose blood-nerve interface [1, 27]. DRG and SDH are immersed in cerebrospinal fluid and there is also neural pathway crosstalk between them. Moreover, CXCR4 is constructively expressed in DRG and SDH neurons [28]. In view of the above, we speculate that plasma CXCL12 may directly enhance the excitability of DRG neurons mainly through high permeability of DRG. Our subsequent results indicated that i.v. injection of CXCL12 significantly increased the excitability of DRG and SDH neurons (Fig. 3B-G) and structural and functional enhancement of nociceptor plasticity (Fig. 5), and upregulated the expression of CXCL12/CXCR4 axis in SDH (Fig. 6). More importantly, intrathecal injection of CXCL12 neutralizing antibody or CXCR4 receptor inhibitor AMD3100 suppressed pain hypersensitivity and pathological changes induced by intravenous CXCL12 (Fig. 7). Multiple lines of evidence suggest that intraperitoneal injection (i.p.) or intrathecal administration (i.t.) of anti-CXCL12 neutralizing antibody or AMD3100 has analgesic effects on many pathological pain states, including a variety of neuropathic pain, such as in a rat SNI [2], a rat L5 spinal nerve ligation model [41, 44, 74], chronic compression of dorsal root ganglion model [79]. These findings are consistent with our results. However, intraperitoneal injection is a systemic administration. In addition, although our studies [40] and Ji RR’s [11, 29, 47] have shown that intrathecal injection of 10 μl in 20 ± g mice is an acceptable mode of administration. However, the drug may not only act on the injected local DRG and spinal cord, but may also affect the brain. Furthermore, activated peripheral glia (such as Schwann cells, satellite glial cells and macrophages) mediate the breakdown of the BNB after nerve injury [6]. Therefore, this drug treatment may have a direct effect on the nervous system and be more likely to inhibit the CXCL12/CXCR4 pathway outside of pain synapses or the nervous system. This also reflects the possibility that these treatments may block plasma CXCL12 pathway. Taken together, our study suggests that the pathological elevation of plasma CXCL12 enhances nociceptor plasticity and ultimately mediates pain chronicity through DRG/BNB leakage and upregulation of SDH CXCL12/CXCR4 axis in positive feedback loop. At the same time, we found that intrathecal blocking of CXCL12/CXCR4 pathways did not completely relieve pain hypersensitivity induced by intravenous CXCL12 (Fig. 7A). It suggests that in addition to the positive feedback loop in DRG and spinal neuronal pathways, other mechanisms such as peripheral nerve endings, brain and immune inflammatory responses cannot be ruled out.

Circulatory immune response was significantly activated in the neuropathic pain model

In recent decades, converging researches have also strongly suggested that activation of the immune system is also an important driver of chronic pain progression [4, 31, 52]. Circulatory immunity plays a vital role in in defense against peripheral nerve injury and disease, and therefore has varying degrees of changes in chronic pain. For example, clinical studies have shown that compared with healthy controls, CRPS have significantly higher proportion of inflammatory monocytes (CD14⁺CD16⁺) in the blood and lower levels of the anti-inflammatory cytokine interleukin-10 (IL-10) in plasma [65]. The peripheral blood of patients with neuropathic pain shows an increase in pro-inflammatory cytokines mediated by CD4⁺ T cells [46]. Our previous studies have demonstrated that two pro-inflammatory cytokines, IL-1β and TNF-α, were significantly increased in the plasma of neuropathic pain model SNI mice [21, 64]. Subsequently, we further found that inflammatory monocytes and neutrophils increased to varying degrees in SNI mice, while lymphocytes decreased [50]. Moreover, increased plasma CXCL12 led to neuroinflammation and memory deficits in mice with neuropathic pain by mediating monocyte transmigration into brain perivascular space. The blood different-expressed genes (DEGs) of both patients with complex back pain and mouse SNI neuropathic pain model have neuropathic and inflammatory components, and there is a very significant correlation [59]. We reanalyzed this data and found that the circulatory immune response was significantly activated in SNI-induced neuropathic pain model compared to CFA-induced inflammatory pain (Fig. 8F). The above evidence confirms substantial changes in circulating immunity in chronic pain, but whether the changes are causally related to the chronicity of pain is unclear.

Causes of increased plasma CXCL12

CXCL12 and its receptor CXCR4 are known to be constructively expressed at multiple sites in the peripheral and central nervous systems (PNS and CNS) [61]. Since then, preclinical animal model studies have shown that this pathway is up-regulated in DRG, SDH and anterior cingulate cortex in several neuropathic pain models [2, 44, 48, 71]. Consistent with these findings, our study further confirmed the overexpression of CXCL12 signaling pathway in the ipsilateral primary pain transduction pathway in the mouse SNI model (Fig. 8A-C). In addition, multiple evidences suggest that Waller degeneration and inflammatory response in the distal nerve stumps are necessary for regeneration and repair [66], but are also responsible for dysfunction such as neuropathic pain [7, 15]. As the primary lesion, more than 20 cytokines were differentially expressed in damaged nerve and DRG at different time points after injury [80]. Our protein microarray results of the sciatic nerve stump at 7 d after SNI injury in mice also showed more than a dozen proteins, such as IL-1β, TNF-α, CCL2 were upregulated (Fig. 8D), and a few molecules of the ipsilateral DRG showed obvious changes (Fig. 8E). These data are consistent with previous studies of rapid upregulation of IL-1β and TNF mRNA and protein levels after sciatic nerve injury in mice [58]. However, the absence of IL-1β and TNF-α, or IL-1 type 1 receptor and TNF-1 receptor led to remission of neutrophil infiltration and mechanical hypersensitivity after sciatic nerve injury. There is evidence that in various neuropathic pain models, the damaged nerves and DRGs are manifested by the disruption of the ipsilateral BBB and BNB [39, 57, 63]. In addition, our previous studies confirm that SNI induced high expression of CXCL12 in perivascular macrophages and endothelial cells [50]. Moreover, we reanalyzed RNA-Seq data from a 2019 study on four sites in a mouse model of SNI neuropathic pain and CFA inflammatory pain (GEO: GSE111216) [59], and found that in addition to DRG and SDH, SNI also led to significant immune responses in the blood, such as IL-1β (Fig. 8F). Activated platelets are reported to be a major source of CXCL12 [10].Taken together, these upregulated molecules in the sciatic nerve stump and DRGs tissues caused by SNI spilled into the blood, affected the blood immune response and ultimately led to elevated plasma CXCL12. However, the source of elevated plasma CXCL12 in pathologic pain remains to be further explored.

In conclusion, this study explored the pathological significance and possible mechanism of plasma CXCL12 elevation in pathologic pain. Our findings revealed that a sustained increase in plasma CXCL12 induces enhanced synaptic structural plasticity of spinal nociceptors through a positive feedback mechanism, which is an important determinant of the chronicity of neuropathic pain.